Introdução à Ciência de Dados e Decisão

1. Objetivo: Explorar e aplicar os conhecimentos obtidos nas aulas expostas sobre Análise exploratória de dados com a linguagem R.

II)Parte 1

Exercicío 1

library(readr)
library(janitor)
library(ggplot2)

caminho <- paste0("C:/Users/monal/Downloads/Curso Ciencias de Dados/us_change.rda")

dados <- load(caminho)

us_change <- janitor::clean_names(us_change) %>% 
  tibble::as_tibble()

#ver o começo do DataSet
head(us_change)

#verdetalhes do Dataset
str(us_change)

tibble [198 x 6] (S3: tbl_df/tbl/data.frame)
 $ quarter     : yearquarter[1:198], format: "1970-01-01" "1970-04-01" ...
 $ consumption : num [1:198] 0.619 0.452 0.873 -0.272 1.901 ...
 $ income      : num [1:198] 1.04 1.23 1.59 -0.24 1.98 ...
 $ production  : num [1:198] -2.452 -0.551 -0.359 -2.186 1.91 ...
 $ savings     : num [1:198] 5.3 7.79 7.4 1.17 3.54 ...
 $ unemployment: num [1:198] 0.9 0.5 0.5 0.7 -0.1 ...
 - attr(*, "key")= tibble [1 x 1] (S3: tbl_df/tbl/data.frame)
  ..$ .rows:List of 1
  .. ..$ : int [1:198] 1 2 3 4 5 6 7 8 9 10 ...
 - attr(*, "index")= chr "Quarter"
  ..- attr(*, "ordered")= logi TRUE
 - attr(*, "index2")= chr "Quarter"
 - attr(*, "interval")=List of 12
  ..$ year       : num 0
  ..$ quarter    : num 1
  ..$ month      : num 0
  ..$ week       : num 0
  ..$ day        : num 0
  ..$ hour       : num 0
  ..$ minute     : num 0
  ..$ second     : num 0
  ..$ millisecond: num 0
  ..$ microsecond: num 0
  ..$ nanosecond : num 0
  ..$ unit       : num 0
  ..- attr(*, "class")= chr "interval"

#primeiro tem que filtrar em dois: 1 para maior que 2000 e 2 para menor que 2000

data_filter_1 <- us_change %>%
  dplyr::select("quarter","unemployment") %>% 
  dplyr::filter(quarter >= "2000-01-01")

data_filter_2 <- us_change %>% 
  dplyr::select("quarter","unemployment") %>% 
  dplyr::filter(quarter < "2000-01-01")

#transformar o primeiro
data.nivel1 <- data_filter_1

data.nivel1$unemployment[1] <-  100; 
for(i in 2:length(data.nivel1$unemployment)){
  data.nivel1$unemployment[i] <-  (1 + data_filter_1$unemployment[i]/100) * data.nivel1$unemployment[i-1]
}

#salvar o primeiro valor
start <- data.nivel1$unemployment[1]

#transformar o segundo
data.nivel2 <- data_filter_2

data.nivel2$unemployment[1] <- (1 + data.nivel2$unemployment[1]/100)* start
for(i in 2:length(data.nivel2$unemployment)){
  data.nivel2$unemployment[i] <-  (1 + data_filter_2$unemployment[i]/100) * data.nivel2$unemployment[i-1]
}

#Agora juntando os datas set
data <- data.nivel2 %>% 
  dplyr::full_join(data.nivel1) %>% 
  tibble::as_tibble()

Joining, by = c("quarter", "unemployment")

Letra B

correl <-  cor(us_change %>% 
      select(-"quarter")) %>% round(2) 

print(correl)

             consumption income production savings unemployment
consumption         1.00   0.38       0.53   -0.26        -0.53
income              0.38   1.00       0.27    0.72        -0.22
production          0.53   0.27       1.00   -0.06        -0.77
savings            -0.26   0.72      -0.06    1.00         0.11
unemployment       -0.53  -0.22      -0.77    0.11         1.00

corrplot::corrplot(correl, 
                   type = "upper",
                   tl.col = "black",
                   )

correl1 <- cor(us_change_nivel %>% dplyr::select(-"quarter")) %>% round(2)

print(correl1)

             consumption income production savings unemployment
consumption         1.00   0.38       0.53   -0.26        -0.03
income              0.38   1.00       0.27    0.72        -0.07
production          0.53   0.27       1.00   -0.06        -0.07
savings            -0.26   0.72      -0.06    1.00        -0.01
unemployment       -0.03  -0.07      -0.07   -0.01         1.00

corrplot::corrplot(correl1, 
                   type = "upper",
                   tl.col = "black",
                   )

Sim, apresentam diferenças porque no primeiro temos o valor “unemployment” em taxa de variação e os outros dados em números, “atrapalhando” a correlação entre os valores e alterando a análise. No segundo, temos todas as colunas na mesma unidade, logo a análise fica confiável.

Letra C

grafico_data <- data %>% 
  ggplot(aes(x = quarter, y = unemployment)) +
  geom_line() +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Evolução do desemprego",
       y = "Quantidade",
       x = "Anos",
       colour = "quarter",
       caption = "Gapminder")

  plotly::ggplotly(grafico_data)

NA

#Gráfico 1

grafico_data1 <- data %>% 
  dplyr::filter(unemployment == max(unemployment)) %>% 
  ggplot(aes(x = quarter, y = unemployment)) +
  geom_point() +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Maior taxa de desemprego",
       y = "Quantidade",
       x = "Anos",
       colour = "quarter",
       caption = "Gapminder")
  
  plotly::ggplotly(grafico_data1)

NA
NA

#Gráfico 2

grafico_data2 <- data %>% 
  ggplot(aes(x = unemployment)) +
  geom_histogram() +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Evolução do Numero de Indice de desemprego",
       y = "Quantidade",
       x = "Número Indice",
       colour = "quarter",
       caption = "Gapminder")
  
  plotly::ggplotly(grafico_data2)

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

#Gráfico 3

grafico_data3 <- us_change %>% 
  ggplot(aes(x = quarter, y=unemployment)) +
  geom_line() +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Evolução da Taxa de desemprego",
       y = "Quantidade",
       x = "Anos",
       colour = "quarter",
       caption = "Gapminder")
  
  plotly::ggplotly(grafico_data3)

NA

Letra D

Observando os gráficos é possível visualizar a alta variação existe ao longo dos anos. Observa-se que houve uma máxima taxa de desemprego no gráfico 4, em outubro. Existe uma grande taxa de variação como demonstra o gráfico de variação. Bem como também tem-se grande variação no número índice. No histograma tem-se a quantidade de desemprego por número índice ao longo dos anos.

Letra E

Sim, várias mudanças brucas ao longo do tempo. Temos o periodo de 1980 até 2000, onde o indice de desemprego foi maior registrado e próximo do ano 2000 tem-se o menor índice.

Exercício 2

Letra A - O cuidado adicional inserido na importação do arquivo Retail.xlsx foi o (ship=1) para pular uma linha,porque a planilha do Excel possui duas linhas de cabeçalho.

dados_2 <- readxl::read_excel(path = "C:/Users/monal/Downloads/Curso Ciencias de Dados/retail.xlsx",skip=1) %>% 
  janitor::clean_names() %>%
  tibble::as_tibble()

Letra B - Para transformar o tipo POSIXct na coluna Series_id utilizou-se a biblioteca “lubridate” com a função (as.date())

dados_2 <- dados_2 %>% 
  dplyr::mutate(
    series_id = lubridate::as_date(series_id)
  )

Letra C e D - Essa questão foram feitas juntas de forma a observar a sazonalidade e a têndencia bem como também o outlier ou padrões anômalos. Decompondo pela série utilizando o método X11, observou-se datas especificas de picos de vendas analisada no data set importado.

  
dados2.ts <- dados_2 %>% 
  dplyr::select(c("a3349335t")) %>% 
  ts(
    start = c(
      lubridate::year(dplyr::first(dados_2$series_id)),
      lubridate::month(dplyr::first(dados_2$series_id))),
    
    end = c(
      lubridate::year(dplyr::last(dados_2$series_id)), 
      lubridate::month(dplyr::last(dados_2$series_id))),
    
    frequency = 12
  ) %>% 
  .[,1]

grafico_1 <- autoplot(dados2.ts) + ggtitle("A3349335T") + 
  theme(plot.title = element_text(hjust = 0.5)) + 
  xlab("Anos") +
  ylab("Total de vendas por Dolar$'000s")
  
grafico_2 <- ggseasonplot(dados2.ts, year.labels=TRUE, year.labels.left=TRUE) +
  theme(plot.title = element_text(hjust = 0.5)) +  
  ylab("Total de vendas em dolar$'000s por Ano") +
  ggtitle("Evolução por estação : A3349335T")
  
grafico_3 <- ggmonthplot(dados2.ts) +
  theme(plot.title = element_text(hjust = 0.5)) + 
  ylab("Total de vendas em dolar$'000s") +
  ggtitle("Evolução por Mês: A3349335T")
    
dados2ts2 <- window(dados2.ts, start=1982)

'start' value not changed

grafico_4 <- gglagplot(dados2ts2) + 
  ggtitle("Venda mensal") +
  theme(plot.title = element_text(hjust = 0.5))
  

decomposicao <- stl(dados2.ts, s.window = "periodic", robust = TRUE)  %>%
  pluck("time.series") %>% 
  as_tibble() %>% 
  mutate(
    date = seq(from = lubridate::as_date(dplyr::first(dados_2$series_id)),
               to = lubridate::as_date(dplyr::last(dados_2$series_id)),
               by = "months")
  ) %>% 
  pivot_longer(
    cols = -"date",
    names_to = "decomposition",
    values_to = "valores"
  )
 
  
grafico_5 <- decomposicao %>% 
  filter(decomposition == "seasonal") %>% 
  ggplot(aes(x = date, y = valores)) + 
  theme(plot.title = element_text(hjust = 0.5)) +
  geom_line()+ 
  labs(title = "Sazonalidade",
       y = "Quantidade",
       x = "Anos",
       caption = "Gapminder")

grafico_6 <-  decomposicao %>% 
  filter(decomposition == "trend")  %>% 
  ggplot(aes(x = date, y = valores)) + 
  theme(plot.title = element_text(hjust = 0.5)) +
  geom_line() + 
  labs(title = "Tendência",
       y = "Quantidade",
       x = "Anos",
       caption = "Gapminder")
  
outlier <- dados_2 %>% 
  anomalize::time_decompose("a3349335t") %>%
  anomalize::anomalize(remainder) %>%
  anomalize::time_recompose() %>% 
  dplyr::filter(anomaly == "Yes") %>% 
  dplyr::select(series_id) 

Converting from tbl_df to tbl_time.
Auto-index message: index = series_id
frequency = 12 months
trend = 60 months

#Analise dos Dados

plotly::ggplotly(grafico_1)

#GGseasonplot

plotly::ggplotly(grafico_2)

#ggmonthplot

plotly::ggplotly(grafico_3)

#Exploração dos dados

plotly::ggplotly(grafico_4)

Aspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/widthAspect ratios aren't yet implemented, but you can manually set a suitable height/width

#Sazonalidade

plotly::ggplotly(grafico_5)

#Tendencia

plotly::ggplotly(grafico_6)

outlier

# A time tibble: 8 x 1
# Index: series_id
  series_id 
  <date>    
1 2007-12-01
2 2008-12-01
3 2009-11-01
4 2009-12-01
5 2010-12-01
6 2011-12-01
7 2012-12-01
8 2013-12-01

Parte - 2

Exercício 1 - Spotify

library(gapminder)
library(tidyverse)
library(janitor)


songs <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-01-21/spotify_songs.csv')

Parsed with column specification:
cols(
  .default = col_double(),
  track_id = col_character(),
  track_name = col_character(),
  track_artist = col_character(),
  track_album_id = col_character(),
  track_album_name = col_character(),
  track_album_release_date = col_character(),
  playlist_name = col_character(),
  playlist_id = col_character(),
  playlist_genre = col_character(),
  playlist_subgenre = col_character()
)
See spec(...) for full column specifications.

songs<- songs %>% 
  janitor::clean_names() %>% 
  tibble::as_tibble()


#Analisando dos dados

data_songs_artist <- songs %>% 
  dplyr::select("energy","track_popularity","track_artist","track_album_release_date")


grafic_1 <- data_songs_artist %>% 
  dplyr::filter(track_popularity > 90) %>% 
  dplyr::filter(track_album_release_date >= "2019-06-01") %>% 
  ggplot(aes(x = track_album_release_date, y = track_popularity)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  geom_point() +
  labs(title = "Popularidade (maior 90) entre os artistas ", 
       subtitle = "Por data",
       y = "Popularidade",
       x = "Data",
       colour = "track_artist",
       caption = "Gapminder")
  

grafic_2 <- data_songs_artist %>% 
  dplyr::filter(track_popularity > 90) %>% 
  dplyr::filter(energy > 0.7) %>% 
  ggplot(aes(x = energy, y=track_popularity)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  geom_line() +
  labs(title = "Popularidade (maior 90) entre os artistas ", 
       subtitle = "Por energia (maior 0.7)",
       y = "Popularidade",
       x = "Energy",
       colour = "track_artist",
       caption = "Gapminder")
  
grafic_3 <- data_songs_artist %>% 
  dplyr::filter(energy == max(energy)) %>% 
  ggplot(aes(x = track_popularity, y = energy)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  geom_point() +
  labs(title = "Maior Energy ", 
       subtitle = "Por energia",
       y = "Energy",
       x = "Popularidade",
       colour = "track_artist",
       caption = "Gapminder")
  
grafic_4 <- data_songs_artist %>% 
  ggplot() +
  geom_histogram(aes(energy)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Quantidade de musica", 
       subtitle = "Por energia",
       y = "Count music",
       x = "Energy",
       caption = "Gapminder")

songs <- songs %>% 
  dplyr::mutate(track_album_release_date = lubridate::as_date(track_album_release_date))

 1886 failed to parse.

  

data_songs <- songs %>% 
  dplyr::select("track_popularity","track_album_release_date",
                "track_album_name")

grafic_5 <- data_songs %>% 
  dplyr::filter(track_popularity > 80) %>% 
  ggplot(aes(x = track_album_release_date, y= track_popularity)) +
  geom_line()+
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Popularidade entre os artistas ", 
       subtitle = "Por data de lançamento",
       y = "Popularidade",
       x = "Data de Lançamento",
       colour = "track_album_name",
       caption = "Gapminder")

#Analise 1- Spotify

plotly::ggplotly(grafic_1)

#Analise 2 -Spotify

plotly::ggplotly(grafic_2)

#Analise 3 -Spotify

plotly::ggplotly(grafic_3)

#Analise 4 -Spotify

plotly::ggplotly(grafic_4)

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

#Analise 5 - Spotify

plotly::ggplotly(grafic_5)

Exercício 2 - Video Games

library(tidyverse)
library(lubridate)

#importanto o DataFrame
video_games <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-07-30/video_games.csv")

Parsed with column specification:
cols(
  number = col_double(),
  game = col_character(),
  release_date = col_character(),
  price = col_double(),
  owners = col_character(),
  developer = col_character(),
  publisher = col_character(),
  average_playtime = col_double(),
  median_playtime = col_double(),
  metascore = col_double()
)

head(video_games)

video_games <- video_games %>% 
  janitor::clean_names() %>% 
  dplyr::mutate(
    release_date = mdy(release_date)
  )

 1 failed to parse.

data_filter_price <- video_games %>% 
  dplyr::select("release_date","price") %>% 
  dplyr::filter(price < 0.99)

graf_1 <- data_filter_price %>% 
  ggplot() +
  geom_line(aes(x = price, y = release_date)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Comparação dos preços dos jogos ", 
       subtitle = "Por data",
       y = "Preço (< 0,99)",
       x = "Ano",
       colour = "price",
       caption = "Gapminder")
 

graf_2 <- data_filter_price %>%
  dplyr::filter(price == max(price)) %>% 
  ggplot() +
  geom_point(aes(x = release_date, y = price)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Evolução dos preços dos jogos", 
       subtitle = "Por data",
       y = "Preço (em dolares)",
       x = "Ano",
       caption = "Gapminder")

graf_3 <- data_filter_price %>%
  ggplot() +
  geom_histogram(aes(price)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Quantidade de jogos por preço", 
       subtitle = "Por preço (Price < 0.99)",
       x = "Preço (em dolares)",
       y = "Count games",
       caption = "Gapminder")


data_filter_game <- video_games %>% 
  dplyr::select("game","metascore") %>%
  janitor::clean_names() %>% 
  tibble::as_tibble() %>% 
  tidyr::drop_na() 

graf_4 <- data_filter_game %>% 
  ggplot() +
  geom_histogram(aes(metascore)) +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Metascore", 
       subtitle = "Por game",
       x = "Metascore",
       y = "Count games",
       caption = "Gapminder")

#Análise 1 - Video Games

 plotly::ggplotly(graf_1)

#Análise 2 - Video Games

plotly::ggplotly(graf_2)

#Análise 3 - Video Games

plotly::ggplotly(graf_3)

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

#Análise 4 - Video Games

plotly::ggplotly(graf_4)

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.